NetMHC 4.0

Overview

NetMHC 4.0 is a computational tool for predicting peptide binding affinity to MHC class I molecules (HLA alleles) using artificial neural networks trained on experimentally measured binding data. It is widely used for neoantigen prediction: candidate peptides (e.g., from somatic mutations or fusion junctions) are scored for binding to the patient’s HLA type, and those predicted to bind strongly are nominated as potential neoantigens for T-cell recognition.

Used by

• Used for HLA-typed neoantigen prediction from fusion-junction peptides across 9,624 TCGA tumor samples; mean 1.5 predicted neoantigens per fusion (range 0.33 in KICH to 2.88 in THYM); frameshift fusions yielded more epitopes than inframe fusions (mean 2.2 vs. 1.0); results described as “exploratory and speculative” PMID:29617662

Notes

• Neoantigen predictions are theoretical; experimental validation (MHC binding assays, T-cell response assays) is required to confirm immunogenicity.
• Frameshift fusions generate longer novel peptide sequences and thus more neoantigen candidates than inframe fusions, though the former may be subject to nonsense-mediated decay.
• HLA typing is required as input; TCGA HLA types were available for the cohort.

Sources

This page was processed by crosslinker on 2026-05-15.